Identification of key immune-related genes and potential therapeutic drugs in diabetic nephropathy based on machine learning algorithms.

Chang Guo; Wei Wang; Ying Dong; Yubing Han

doi:10.1186/s12920-024-01995-4

Home/Archive/PubMed

PubMed · 39187837

Identification of key immune-related genes and potential therapeutic drugs in diabetic nephropathy based on machine learning algorithms.

Chang Guo·Wei Wang·Ying Dong·Yubing Han

View original source ↗Export metadata ↓

Abstract

BACKGROUND: Diabetic nephropathy (DN) is a major contributor to chronic kidney disease. This study aims to identify immune biomarkers and potential therapeutic drugs in DN. METHODS: We analyzed two DN microarray datasets (GSE96804 and GSE30528) for differentially expressed genes (DEGs) using the Limma package, overlapping them with immune-related genes from ImmPort and InnateDB. LASSO regression, SVM-RFE, and random forest analysis identified four hub genes (EGF, PLTP, RGS2, PTGDS) as proficient predictors of DN. The model achieved an AUC of 0.995 and was validated on GSE142025. Single-cell RNA data (GSE183276) revealed increased hub gene expression in epithelial cells. CIBERSORT analysis showed differences in immune cell proportions between DN patients and controls, with the hub genes correlating positively with neutrophil infiltration. Molecular docking identified potential drugs: cysteamine, eltrombopag, and DMSO. And qPCR and western blot assays were used to confirm the expressions of the four hub genes. RESULTS: Analysis found 95 and 88 distinctively expressed immune genes in the two DN datasets, with 14 consistently differentially expressed immune-related genes. After machine learning algorithms, EGF, PLTP, RGS2, PTGDS were identified as the immune-related hub genes associated with DN. In addition, the mRNA and protein levels of them were obviously elevated in HK-2 cells treated with glucose for 24 h, as well as their mRNA expressions in kidney tissues of mice with DN. CONCLUSION: This study identified 4 hub immune-related genes (EGF, PLTP, RGS2, PTGDS), as well as their expression profiles and the correlation with immune cell infiltration in DN.

Explore related subjects

Diabetic Nephropathies Machine Learning Humans Animals Mice RGS Proteins Gene Expression Profiling Molecular Docking Simulation Algorithms Diabetic nephropathy Diagnostic biomarkers Immune-related genes Machine learning Journal Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Chang Guo, Wei Wang, Ying Dong, Yubing Han. 2024-08-26. Identification of key immune-related genes and potential therapeutic drugs in diabetic nephropathy based on machine learning algorithms.. https://doi.org/10.1186/s12920-024-01995-4

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2026

Identification and expression validation of key genes of Xiaozhengtongluo formula in the treatment of diabetic nephropathy by Mendelian randomization.

Xiaozhengtongluo formula (XZTL) has a positive effect on the treatment of diabetic nephropathy (DN), but its mechanism is not fully understood. Therefore, it is important to explore the key genes of XZTL in the treatment of DN. Differentially expressed genes (DEGs) between DN and control obtained from GSE96804, drug target genes of XZTL, and disease target genes of DN obtained from public databases were intersected. Genes of intersection were defined as candidate genes. Next, Mendelian randomization (MR) analysis was used to ascertain the causal associations between candidate genes and DN. Afterwards, key genes were confirmed through receiver operating characteristic (ROC) curve analysis and expression validation. Subsequently, enrichment analysis, molecular regulatory network analysis, and molecular docking were conducted. Finally, experimental verification of the expression levels of key genes was performed through reverse transcription-quantitative polymerase chain reaction (RT-qPCR). Altogether, 29 candidate genes were screened via MR analysis, identifying APOD, IGFBP3, and LPL as significantly associated with DN. IGFBP3 and APOD were risk factors, whereas LPL was protective. Consistent expression trends across training and validation datasets defined them as key genes. All three were co-enriched in 26 pathways, including oxidative phosphorylation. Regulatory networks showed MIR497HG/hsa-miR-19a-3p regulated IGFBP3, and NEAT1/hsa-miR-29a-3p regulated LPL; IGFBP3 and LPL were co-targeted by SP3 and SP1. Molecular docking revealed APOD-baicalein, LPL-oleic acid, and IGFBP3-quercetin binding, suggesting therapeutic potential. RT-qPCR confirmed aberrant expression of these genes in DN, which was normalized by XZTL intervention. In this study, three key genes (APOD, IGFBP3, and LPL) of XZTL in the treatment of DN were finally obtained, providing mechanistic clues for understanding XZTL's multi-target mechanism and providing experimentally tractable candidate targets for DN molecular subtyping, targeted therapeutic development, and precision medicine approaches in TCM.

Diabetic Nephropathies↗

PubMed2025

Genomic mapping of diabetic kidney disease biomarkers and identification of potential inhibitors through virtual screening.

BACKGROUND: Diabetic kidney disease (DKD) is a common and serious complication of diabetes mellitus, marked by a multifactorial pathogenesis and the absence of sensitive diagnostic biomarkers. Identifying novel molecular targets and therapeutic options is essential to improve early diagnosis and treatment outcomes. METHODS: To uncover potential biomarkers and therapeutic candidates, we performed an integrated genomic analysis using microarray and RNA-seq datasets from the Gene Expression Omnibus (GEO) and Sequence Read Archive (SRA) databases. Differentially expressed genes (DEGs) were identified and subjected to protein-protein interaction (PPI) network analysis. Key genes were further explored through virtual screening of an FDA-approved compound library using molecular docking techniques. Drug-likeness was assessed via Lipinski's rule of five. RESULTS: A total of 40 DEGs were identified, among which ISCU (downregulated; involved in iron-sulfur cluster biogenesis) and AP1S2 (upregulated; associated with vesicular trafficking) emerged as potential biomarkers. PPI analysis revealed their involvement in critical DKD-related pathways, such as extracellular matrix remodeling and oxidative stress. Virtual screening identified six FDA-approved compounds with high binding affinity (≤-7.96 kcal/mol) to ISCU, notably ZINC000001576020, all of which complied with Lipinski's rule. CONCLUSIONS: This in-silico study nominates ISCU and AP1S2 as candidate diagnostic biomarkers for DKD and identifies computationally prioritized inhibitors targeting ISCU. These findings require experimental validation but provide a molecular framework for precision diagnosis and therapeutic development. These findings offer new molecular insights that could inform precision diagnosis and personalized treatment strategies for diabetic kidney disease.

Diabetic Nephropathies↗